Sh3Gen was internally too complex and using a huge amount of variables (specific for time, theatre, faction, unit, ..). In the end I was never quite sure how it worked (when it comes to randomization). The good thing was that this made it unpredictable for the user.
Right now the TR 2.0 application works without randomization. I found that it's much easier to develop from that. It' a possibility that I add randomization, but later, and of a much much simpler principle. Not that this should matter to the user. The result should be at least the same, and scale better with changes to the variables. Right I rather try to design the output in a way that it can be used
without heavy randomization.
Look at my screen to see what I mean.
Screen1 shows just the following: an unknown number of merchants ships, as well as some unidentified naval unit is docked in the port. Patrols cover both entrances. There is also information about the position of coastal batteries as well as some coastal traffic.
Screen 2: other scale, same principle. Colored zone shows where japanese merchant traffic is concentrated but it's a large area and you don't see any individual units. There are also a multitude of other zones for coastal traffic, aircoverage etc, but they never reveal any information that's dead on.
[IMG]file:///C:/DOKUME%7E1/Standard/LOKALE%7E1/Temp/moz-screenshot-2.jpg[/IMG]
If you have used Sh3Gen you know how I do things, and that I do not like the spysat approach. But there have been a lot of complaints by people about never being able to find things plotted by Sh3Gen. I hope it makes sense if I therefore try to
raise accuracy first, before anything else.